Systems and Methods for Real-Time Video Enhancement

1. A computer-implemented method for improving live video quality comprising: (a) acquiring, using a medical imaging apparatus, a stream of consecutive image frames of a subject, wherein the stream of consecutive image frames is acquired with a reduced amount of radiation dose; (b) applying a deep learning network model to the stream of consecutive image frames to generate an output image frame with improved quality in both temporal domain and spatial domain, wherein the deep learning network model is trained using training datasets comprising a pair of a simulated low-quality video and a simulated high-quality video; and (c) displaying the output image frame with improved quality in real-time on a display.

2. The computer-implemented method of claim 1, wherein the simulated high-quality video is generated by applying a temporal averaging algorithm or a denoising algorithm to a video acquired with a normal radiation dose.

3. The computer-implemented method of claim 2, further comprising computing a noise based on a difference between the video and the simulated high-quality video.

4. The computer-implemented method of claim 2, further comprising applying a factor to the noise to simulate a level of noise corresponding to the factor.

5. The computer-implemented method of claim 4, wherein the simulated low-quality video is generated based at least in part on the level of noise and the simulated high-quality video.

6. The computer-implemented method of claim 1, wherein the deep learning network model comprises a plurality of denoising components.

7. The computer-implemented method of claim 6, wherein the plurality of denoising components are assembled in a two-layer architecture.

8. The computer-implemented method of claim 7, wherein each denoising component in a first layer of the two-layer architecture processes a subset of the stream of consecutive frames to output a series of intermediate image frames with an enhanced image quality.

9. The computer-implemented method of claim 8, wherein a denoising component in the second layer of the two-layer architecture processes the intermediate image frames with the enhanced image quality and generates the output image frame.

10. The computer-implemented method of claim 7, wherein each denoising component includes a modified U-net model.

11. The computer-implemented method of claim 10, wherein a denoising component in a second layer of the two-layer architecture has weights different from the weights of a denoising component in the first layer.

12. The computer-implemented method of claim 1, wherein the medical imaging apparatus is performing fluoroscopic imaging.

13. A non-transitory computer-readable storage medium including instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising: (a) acquiring, using a medical imaging apparatus, a stream of consecutive image frames of a subject, wherein the stream of consecutive image frames is acquired with a reduced amount of radiation dose; (b) applying a deep learning network model to the stream of consecutive image frames to generate an output image frame with an improved quality in both temporal domain and spatial domain, wherein the deep learning network model is trained using training datasets comprising a pair of a simulated low-quality video and a simulated high-quality video; and (c) displaying the output image frame with the improved quality in real-time on a display.

14. The non-transitory computer-readable storage medium of claim 13, wherein the simulated high-quality video is generated by applying a temporal averaging algorithm or a denoising algorithm to a video acquired with a normal radiation dose.

15. The non-transitory computer-readable storage medium of claim 14, wherein the one or more operations further comprise computing a noise based on a difference between the video and the simulated high-quality video.

16. The non-transitory computer-readable storage medium of claim 14, wherein the one or more operations further comprise applying a factor to the noise to simulate a level of noise corresponding to the factor.

17. The non-transitory computer-readable storage medium of claim 16, wherein the simulated low-quality video is generated based at least in part on the level of noise and the simulated high-quality video.

18. The non-transitory computer-readable storage medium of claim 13, wherein the deep learning network model comprises a plurality of denoising components.